Physicochemical Characterization of the Orthorhombic Polymorph ofParacetamol Crystallized from Solution

GARY NICHOLS*,† AND CHRISTOPHER S. FRAMPTON‡

Contribution from Pfizer Central Research, Ramsgate Road, Sandwich, Kent CT13 9NJ, England and Roche DiscoveryWelwyn, Broadwater Road, Welwyn Garden City, Hertfordshire AL7 3AY, England.

Received December 30, 1997. Accepted for publication March 9, 1998.

Abstract 0 This paper describes a method for the laboratory-scalecrystallization of the orthorhombic polymorph (form II) of paracetamol(acetaminophen) from solution. Its structure has been determinedby single-crystal X-ray crystallography at 298 K (to confirm the resultsof data published in 1974) and at 123 K (to improve the overallaccuracy of the structure determination). Despite considerable effortby many investigators, the crystallization of form II from solution, usingthe method given in the 1974 structure report, has been elusive. Theincentive for this effort is that form II, unlike commercial paracetamol(form I), undergoes plastic deformation and is suitable for directcompression. Consequently, the ability to produce form II in quantityhas attracted much interest because of the potential commercialbenefits to be gained by not using binders during the manufacture oftablets. However, until now, the only method that has been reportedfor the bulk preparation of form II has been to grow it as polycrystallinematerial from fused form I. This study also compares the solid-stateproperties of form II with those of form I, with particular emphasis onthe crystallography (both X-ray and optical), crystal morphology, thermalbehavior, and compaction properties.

IntroductionThis paper reports the results of an investigation that

was initiated by the repeated failure of researchers tocrystallize the orthorhombic polymorph (form II) of parac-etamol from solution using the method described in 1974by Haisa et al.1 Since the initial results of this currentwork were first published,2 the potential commercialsignificance of the findings became evident. Therefore,additional studies have subsequently been conducted (1)to develop a laboratory-scale crystallization method and(2) to characterize the physicochemical properties of formII in comparison with those of form I.Paracetamol (acetaminophen) is an analgesic drug that

is used worldwide in the manufacture of many millions oftablets and other dosage forms every year. The crystalstructures for the two known crystal modifications ofparacetamol (monoclinic and orthorhombic) have alreadybeen published.1,3 Monoclinic paracetamol is the thermo-dynamically stable modification at room temperature withrespect to the orthorhombic modification. Evidence hasalso been published which indicates that a third modifica-tion exists.4,5 However, this third polymorph (form III) hasonly been observed during fusion experiments and isreported to be so unstable that, as yet, no crystals havebeen isolated to enable its structure or physicochemicalproperties to be determined.

Monoclinic paracetamol (form I), which is the com-mercially used form, is not suitable for direct compressioninto tablets. This is because it lacks slip planes in itscrystal structure, which are a prerequisite for plasticdeformation upon compaction. Consequently, form I hasto be mixed with binding agents before tabletting, whichis costly in both time and materials. However, directlycompressible form I has been prepared by modifying thecrystallization process; for example, by the desolvation ofa hemisolvate to produce sintered-like crystals6,7 and bycrystallization in the presence of 0.5% of PVP 10000 or50000.8 In contrast, orthorhombic paracetamol (form II)has well-developed slip planes in its crystal structure and,as a result, it undergoes plastic deformation upon compac-tion.9 For this reason, it has been postulated that theorthorhombic form of paracetamol may have distinctprocessing advantages over the monoclinic form.9 Inaddition to the commercial benefit of direct compression,there is evidence to suggest that form II, which is themetastable polymorph, may also be slightly more solublethan form I.10

Despite considerable efforts to crystallize the ortho-rhombic polymorph by slow evaporation from an ethanolsolution, as described by Haisa et al.,1 this method ofproducing it has eluded subsequent researchers.5,10 Theonly method that has been reported for the reproduciblebulk production of polymorphically pure orthorhombicparacetamol powder is by crystallization from meltedmonoclinic paracetamol in a nonoxidizing atmosphere.5Consequently, studies into the pharmaceutical and me-chanical properties of form II have been performed usingmilled, polycrystalline material. However, crystallizationof form II from a solution would be a more desirable andcontrollable manufacturing process for industrial scale-uppurposes. Monoclinic paracetamol, on the other hand, isreadily produced from aqueous solution3 and many othersolvents.10,11

To crystallize form II from solution, a supersaturatedsolution of paracetamol was nucleated with seeds of formII (from melt-crystallized paracetamol) while the crystal-lization was observed by light microscopy. The crystal-lization of form II was successful and this paper describesa laboratory-scale method that has been developed fromthe microscale to crystallize orthorhombic paracetamolfrom solution. The identity of form II has been confirmedby single-crystal X-ray analysis and powder X-ray diffrac-tion. Single-crystal structure analyses of both forms I andII have been determined at 298 K (+25 °C) to confirm theresults of Haisa et al.1,3 and at 123 K (-150 °C) to improvethe overall accuracy of the structure determination. Theresults of the 123 K data collections are presented herebecause these supplement and enhance the existing struc-tural data. The unit cell parameters for both forms I andII at room temperature are also reported and are comparedwith previously published data.

* Corresponding author. Telephone number: (0) 1304 616161. Faxnumber: (0) 1304 623909. E-mail address: gary_nichols@sandwich.pfizer.com.

† Pfizer Central Research.‡ Roche Discovery Welwyn.

S0022-3549(97)00483-8 CCC: $15.00684 / Journal of Pharmaceutical Sciences © 1998, American Chemical Society andVol. 87, No. 6, June 1998 American Pharmaceutical AssociationPublished on Web 05/06/1998

In addition to the confirmation of the crystal structuresof forms I and II, the morphologies of forms I and II havebeen simulated and the predicted models are comparedwith the morphologies of the experimentally grown crys-tals. Also, the compaction properties of forms I and II (bothcrystallized from solution) have been determined using acompaction simulator. To our knowledge, the diagnosticoptical properties of form II have never been fully charac-terized and published. For completeness, therefore, theprincipal refractive indices for form II were determined andcompared with those for form I.

Experimental Section

MaterialssParacetamol (4-acetamidophenol) (Sigma ChemicalCo., Poole, Dorset, England) was used for the preparation of formII. This material was confirmed as being form I by powder X-raydiffraction. The solvent used to crystallize form II was BP (BritishPharmacopeia) grade industrial methylated spirits (IMS) whichis ethanol with approximately 4% methanol (by volume). Benzylalcohol (Aldrich Chemical Co., Gillingham, Dorset, England) wasused for the preparation of single crystals of form II.Preparation of Form II Seed CrystalssPolycrystalline form

II is readily prepared by the fusion of form I.10 About 5 mg ofform I was melted on a glass microscope slide over a spirit lamp.The melt was vitrified by rapid cooling to room temperature afterplacing the slide on an aluminum block. Within 30 min of cooling,the glass had started to crystallize as form II. Seeds were collectedby gently scraping the top surface of the crystallized melt with amicrospatula. Due to the simplicity of this method to grow seedsof pure form II, new seeds were grown for every solvent crystal-lization experiment.Crystallization of Orthorhombic Paracetamol from Solu-

tionsTwo different methods were used to crystallize form II forthe single-crystal X-ray analysis (Method 1) and for all otherexperiments (Method 2).Method 1sFor the isolation of well-formed crystals that are

suitable for single-crystal X-ray analysis, crystallization from asupersaturated, high-boiling point solvent has been suggested byMcCrone.12 Benzyl alcohol, which boils at 205 °C, has been usedsuccessfully to grow crystals of form II using the followingmicrocrystallization method, which requires the use of a transmit-ted light compound microscope (magnification of about 40×) toobserve the development of the crystals and to enable the singlecrystals to be manipulated and harvested. A single drop of abenzyl alcohol (about 1 mL) on a microscope slide was warmed toabout 50 °C on a hot plate and paracetamol was added until nomore would dissolve. The drop was then warmed to about 80 °Cto ensure complete dissolution of the paracetamol. The microscopeslide was then rapidly chilled, to achieve a supersaturated solution,by placing it on an aluminum block that had been prechilled to 0°C on ice. The chilled drop was seeded with a few micrograms ofform II (prepared from a recrystallized melt) and was leftuncovered. The microscope slide was maintained at 0 °C and wasexamined periodically, using the light microscope, to monitor thegrowth of needle-shaped form II crystals. After about 15 min, thelargest crystals (between 100 and 200 µm long) were isolated fromthe drop using a steel needle probe and dried with a piece of filterpaper that had been cut to a sharp point.Method 2sAttempts to crystallize form II by cooling supersatu-

rated IMS solutions (without seeding) at different temperatures(22 °C, 4 °C, -34 °C, and -75 °C) always resulted in the growthof form I. However, as described later, subambient temperaturesdid sometimes favor the growth of form II, provided that thecrystals were harvested soon after the onset of crystallization.The next approach was to nucleate the supersaturated solution

using seeds of form II that had been prepared from a crystallizedmelt. This nucleation method was successful, so a simple andrepeatable method for the laboratory-scale preparation of ortho-rhombic paracetamol from IMS was developed, as described below.The amounts of paracetamol (form I) needed to give saturated

solutions in IMS at room temperature (22 °C) and at 0 °C werefound to be 6.75 g in 50 mL (0.135 g/mL) and 5.00 g in 50 mL (0.1g/mL), respectively. These values were used as a guide to preparesupersaturated solutions for the crystallization experiments.

A supersaturated solution of paracetamol was prepared bydissolving 9.3 g of form I in 50 mL of IMS at about 50 °C in astoppered 150 mL conical flask. This solution was agitated untilthe paracetamol was completely dissolved. The warm solution wascarefully transferred (using a funnel to prevent splashing) into asecond, clean, dry, 150 mL conical flask that had been prechilledin an ice bath at 0 °C. After cooling for about 10 min, the solutionwas seeded with a small amount (about 50 µg) of the powdered,melt-crystallized form II.The seeded solution was left to stand at 0 °C, with occasional

mixing by gentle swirling. Precipitation and growth of form IIcrystals occurred rapidly (usually within 10 min of seeding). After15-20 min from seeding, the precipitated crystals of form II wereharvested by rapid suction filtration onto a paper filter and thenair-dried. It is very important that the form II crystals arecompletely dried as quickly as possible. This is because residualsolvent can induce a partial conversion of form II to form I duringstorage (via a solvent-mediated transformation as discussed later).It was observed that if the precipitation was allowed to continue

for more than a few minutes (in an attempt to increase the yieldof form II), crystals of form I began to grow. For the purpose ofthese laboratory-scale experiments, the yield was not of primeimportance. Generally, the yield of pure form II was low (typicallyless than 30%); therefore, to increase the yield for commercialproduction of form II, the crystallization and recovery processwould need to be optimized.Analysis and CharacterizationsA combination of analytical

techniques was used to characterize and compare the physico-chemical properties of the orthorhombic and monoclinic poly-morphs of paracetamol.Single-Crystal X-ray DiffractionsSingle crystals of form I were

grown from an ethanol solution and crystals of form II were grownfrom benzyl alcohol. Diffraction data were collected at tempera-tures of 298 and 123 K on a Rigaku AFC7R diffractometerequipped with a graphite monochromator, using Mo KR radiation.An Oxford Cryosystems cryostream cooler13 (Oxford Cryosys-

tems, Long Hanborough, Oxford, England) was used to cool thecrystals to 123 K during the collection of the data. The intensitieswere collected in the ω-2θ scan mode. The recorded reflectionswere corrected for Lorentz and polarization effects. No absorptioncorrection was applied since preliminary ψ-scan measurementsrevealed no significant absorption effects. Three orientation andintensity standards were monitored after every 150 reflectionsrecorded; no significant variation was observed for either sample.The structures were solved by direct methods, followed by full-

matrix least-squares refinement on F2. All non-hydrogen atomswere refined with anisotropic temperature factors. All hydrogenatoms were located in subsequent difference maps and freelyrefined with isotropic temperature factors with convergence at∆/σmax.< 0.005. Programs used for the structure determinationwere teXsan, version 1.6 (Molecular Structure Corp., The Wood-lands, TX) and SHELXTL (Sheldrick, G. M., Program for crystalstructure refinement; University of Gottingen, Germany, 1996).Powder X-ray Diffraction (PXRD)sThe powder X-ray diffraction

patterns for forms I and II were acquired at room temperature ona Siemens D5000 diffractometer using Cu KR radiation (tubeoperated at 40kV, 40mA), a θ-θ goniometer, automatic divergenceand receiving slits, a graphite secondary monochromator, and ascintillation counter. The data were collected over an angularrange from 2° to 55° 2θ in continuous scan mode using a step sizeof 0.02° 2θ and a step time of 5 s.Form I (Sigma) and form II (grown from IMS solution) were

prepared as flat plate specimens using powder that had beensieved, without grinding, to less than 90 µm using a 170 meshscreen. The powders were packed into 12 mm diameter, 0.5 mmdeep cavities cut into polished, zero-background silicon wafers (TheGem Dugout, 1652 Princeton Drive, Pennsylvania State College,PA). All specimens were rotated in their own plane duringanalysis. Silicon powder (approximately 15% w/w) was used asan internal standard to correct for peak displacement.The diffraction data are reported using Cu KR1 (λ ) 1.5406 Å)

after the KR2 component had been stripped using the Siemens Evasoftware.Differential Scanning Calorimetry (DSC)sThe thermal behav-

iors above ambient of forms I and II were recorded using a Perkin-Elmer DSC Series 7 differential scanning calorimeter equippedwith an automatic sample changer. Approximately 3 mg of each

Journal of Pharmaceutical Sciences / 685Vol. 87, No. 6, June 1998

sample was heated at 10 °C/min from 30 to 190 °C in perforated,crimped aluminum pans while being purged with dry nitrogen.ThermomicroscopysThermally induced events were observed

using a Nikon Labophot transmitted light microscope equippedwith a Mettler FP5 controller and FP52 microscope heating stage.A few crystals were scattered dry between a coverglass andmicroscope slide and these were heated at 5 °C/min from roomtemperature to 175 °C.Optical CrystallographysDespite a careful literature search, no

reference to the optical properties of the orthorhombic polymorphof paracetamol could be found. Only one reference for themonoclinic form, giving minimal optical data, was found.14 Con-sequently, the optical properties for both the orthorhombic andmonoclinic polymorphs were determined and are reported in thispaper.A Nikon Optiphot POL polarizing light microscope, fitted with

a rotating specimen stage, was used for the full optical charac-terizations of both forms I and II. Orthoscopic and conoscopicobservations were made at magnifications from 40× to 1000×.Fresh, certified Cargille refractive index liquids (R. P. Cargille Co.,Cedar Grove, NJ) were used to measure the three principal refrac-tive indices (nR, nâ, and nγ) for both polymorphs by the immersionmethod.15 The refractive index values were determined usingmonochromatic light (λ ) 589 nm) and were corrected to 25 °C.Scanning Electron Microscopy (SEM)sAn Amray 1820T scan-

ning electron microscope was used to examine the crystal habitsof forms I and II. The powders were mounted onto 13 mmdiameter aluminum stubs using double-sided adhesive tape andwere sputter-coated with gold. The scanning electron microscopewas operated with a beam accelerating potential of 3kV, andimages were collected in secondary electron mode.Crystal MorphologysCrystal morphology modeling was per-

formed using Cerius2, version 3.0 (Molecular Simulations Ltd.,Cambridge, England) to compare the predicted crystal habits offorms I and II with those grown experimentally. For bothpolymorphs, two simulated models were calculated: (i) the Bra-vais-Friedel and Donnay-Harker (BFDH) model (which is calcu-lated using the crystal lattice geometry) and (ii) the attachmentenergy (AE) model (which is calculated from the energy releasedas a growth slice attaches onto a specific surface on the crystal).16For the attachment energy calculations, the Dreiding 2.21 forcefield was used.Compaction StudiessPrevious studies on the compressibility

of form II powder derived from melt crystallization have shownthat it can be directly compressed without the need for bindingagents.9 To establish if form II grown from solution is also suitablefor direct compression, exploratory controlled-compaction studieswere performed in triplicate using the method of Heckel.17,18 Forcomparison, the compaction properties of form I (grown from IMSsolution) were also determined under the same conditions. Theparticle sizes of form I and form II used for the compactionexperiments were similar and were measured (using light micros-copy) to be less than 100 µm across, typically between 25 and 50µm.The yield pressures of forms I and II were determined using

an ESH compaction simulator (ESH Testing Ltd., Briefly Hill,West Midlands, England). Flat-faced, 8 mm diameter puncheswere used to compact the powders at velocities of 2 and 220 mms-1. The punch and die faces were lubricated with a 5% w/vsuspension of magnesium stearate in methanol. The fill weightswere calculated (using the density values calculated from the X-rayanalyses) from the amount of material required to give compactsthat would theoretically be 3 mm thick at zero porosity. For eachexperiment, a single compaction was provided using a V-shapedprofile.

Results

Single-Crystal X-ray DiffractionsThe molecular struc-tures of forms I and II determined at 123 K are shown inFigures 1 and 2, respectively. Both structures havesimilar, planar conformations and these agree with thestructures reported by Haisa et al.1,3 This low-temperaturedata enhances the previously published results because theoverall accuracy has been improved. A summary of the

single-crystal data acquired at 123 K for both forms I andII is given in Table 1.Form II has been confirmed as orthorhombic and is

assigned to the space group Pbca, which agrees with thedata of Haisa et al.1 Form I has also been confirmed asmonoclinic, but this study has resulted in the choice of theP21/n space group. This is in agreement with the work ofWelton and McCarthy19 and Wilson et al.20 but differs fromthe assignment of the P21/a space group by Haisa et al.1The room-temperature unit cell parameters for forms I

and II are in accord with those determined by Haisa et al.1,3and are summarized in Table 2.Powder X-ray DiffractionsForms I and II have dif-

ferent crystal structures and can be distinguished fromeach other by PXRD. Figures 3 and 4 show the patternsthat have been acquired from forms I and II, respectively.PXRD data for the first 30 peaks for forms I and II, which

have relative integrated intensities (expressed as a per-centage of the strongest peak) of 1% or greater, are listedin Table 3. The peaks for both polymorphs were indexedusing Cerius2. The powder data for form I agrees with thatgiven by Welton and McCarthy.19 In contrast, the PXRDpattern for form II grown from solution is very differentfrom the pattern from melt-crystallized material.5 Thelatter has a very intense main reflection and many weakreflections due to the effects of preferred orientation.

Figure 1sMolecular structure for monoclinic paracetamol (form I) determinedusing single crystal X-ray diffraction at 123 K. Ellipsoids are drawn at the50% probability level.

Figure 2sMolecular structure for orthorhombic paracetamol (form II)determined using single-crystal X-ray diffraction at 123 K. Ellipsoids are drawnat the 50% probability level.

686 / Journal of Pharmaceutical SciencesVol. 87, No. 6, June 1998

Differential Scanning CalorimetrysThe thermo-gram for form I shows a single endothermic event (Figure5). This sharp peak is due to the melting of form I at about171 °C (onset ) 169 °C; enthalpy ) 186 J g-1).Form II gave a thermogram with three endothermic

events, as shown in Figure 6. The first event, which isbroad and weak, occurs over the temperature range 115to 128 °C and is centered at about 122 °C (onset ) 115 °C;enthalpy ) 2 J g-1). The second event is a weak but sharppeak at about 157 °C (onset ) 157 °C; enthalpy ) 1 J g-1).The third endotherm is a strong, sharp peak at about 171°C (onset ) 169 °C; enthalpy ) 185 J g-1).The thermally induced events for form II have been

interpreted as (in order of increasing temperature) a solid-state conversion of form II to form I, followed by the meltingof nonconverted form II, and finally, the melting of form I.This interpretation was confirmed by thermomicroscopy(see below).Interestingly, the thermal behavior of form II that has

been crystallized from solution is different to that of formII crystallized from the melt. Melt-crystallized paracetamolhas a single strong endothermic event, due to melting, atabout 157 °C.5,10 This suggests that melt-crystallized formII is polymorphically pure, whereas solution grown form Imay have a low level of form I present that is not detectedby PXRD. Alternatively, crystals of form II that precipi-tated rapidly from solution may contain many structuraldefects that could promote the thermally induced conver-sion of form II to form I.

This study suggests that DSC is unlikely to be a reliablemethod for the routine quality control of form II grown fromsolution, because the major thermal event is the meltingof form I that results from the thermally induced solid-state conversion of form II to form I.ThermomicroscopysFrom about 60 °C, individual

crystals of form II converted in the solid-state to form I.This solid-state conversion was a slow and gradual processthat was observed in discrete crystals. The conversion didspeed up slightly from 120 to 135 °C. Between 157 and158 °C, residual crystals of form II melted. Where the meltwas in direct contact with those crystals that had al-ready converted to form I, it recrystallized as form I. Meltthat was not in contact with crystals of form I did notrecrystallize and remained molten. Finally, all the form Icrystals melted, without decomposition, between 170 and171 °C.

Table 1sSummary of Results for the Single Crystal X-ray Analyses ofParacetamol Forms I and II Performed at 123 K

form I form II

chemical formula C8H9NO2 C8H9NO2

formula weight 151.16 151.16temperature (K) 123 123wavelength (Å) 0.71069 0.71069crystal system monoclinic orthorhombicspace group P21/n Pbcaunit cell dimensions

a (Å) 7.0941 (12) 17.1657 (12)b (Å) 9.2322 (11) 11.7773 (11)c (Å) 11.6196 (10) 7.212 (2)â ° 97.821 (10) 90.000

volume (Å3) 753.9 (2) 1458.1 (4)no. of reflct used for cell

deter. (deg θ ranges)25 (17.30 e θ e

19.90)°20 (9.55 e θ e

11.25)°Z 4 8Dc (g/cm3) 1.332 1.377µ(Mo KR) (mm-1) 0.097 0.100F(000) 320 640crystal dimen (mm) 0.30 × 0.30 × 0.15 0.28 × 0.25 × 0.15index ranges

h 0−9 0−21k 0−11 0−15l −14 to 14 0−9

θ range for data collect. (deg) 2.83−26.99 2.94−26.99no. of reflct collected 1771 1588no. of unique reflct 1642 1588Rint 0.0143 no equiv reflct

measuredno. with I g 2σ(I) 1423 955no. of parameters 137 137weighting scheme (A, B)awR2 (all data) 0.0975 0.1144R1 (I g 2σ(I)) 0.0333 0.0423extinction coefficient (x)b 0.0189 (43) 0.0027 (7)goodness of fit 1.013 1.020residual density (e Å3) 0.289 and −0.192 0.230 and −0.202

a w-1 ) σ2(Fo2) + (AP)2 + BP, where P ) (Fo2 + 2Fc2)/3. b F*c ) kFc[1 +0.001xFc2λ3/sin(2θ)]-1/4.

Table 2sSummary of the Crystal Morphology, Unit Cell Parameters,and Crystal Optics for Paracetamol Forms I and II Determined atRoom Temperature

form I form II

Crystal Morphologycrystal system monoclinic orthorhombichabit prisms and plates

elongated parallelto {101h}

prisms elongatedparallel to c-axis

cleavage moderate; parallelto 010

perfect; parallelto 001

Unit Cell Parameterstemperature (K) 298 298space group P21/n Pbca

a (Å) 7.106(2) 17.156(6)b (Å) 9.382(3) 11.831(4)c (Å) 11.704(3) 7.405(3)

axial ratio (a:b:c) 0.757:1:1.247 1.450:1:0.626â angle 97.36(2)° 90.00°cell volume (Å3) 774 1503molecules per unit cell (Z) 4 8density (calcd) g/cm3 1.297 1.336

Crystal Opticsrefractive indices (n25D):

nR 1.580(±0.001) 1.491(±0.001)nâ 1.643(±0.001) 1.667(±0.001)nγ 1.704(±0.001) 1.840(±0.002)

extinction inclined and dispersed:γ ∧ 101h ) 36.2°

straight for prisms,and symmetricalin rhomb-shaped(001) cleavagefragments

Figure 3sExperimental powder X-ray diffraction pattern for monoclinicparacetamol (form I).

Journal of Pharmaceutical Sciences / 687Vol. 87, No. 6, June 1998

The solid-state conversion of form II to form I was readilymonitored between crossed polarizers. Form II crystals attheir extinction position (i.e. black when they are parallelwith one of the polarizers), became bright as they convertedto the monoclinic form I (which has inclined extinctionwhen viewed along the b-axis).Optical CrystallographysThe solid-state optical prop-

erties of a specific compound, such as refractive index,birefringence, and extinction angle,15 are controlled by itsmolecular structure and are therefore a physical constantfor that compound. The optical properties for both themonoclinic and orthorhombic forms of paracetamol aregiven in Table 2. The polymorphs of a compound haveunique molecular structures and chemical properties and,as a consequence, each has unique and diagnostic optical

properties. These optical properties can be used to distin-guish between the polymorphs of paracetamol.Comparison of the optical properties for forms I and II

shows that they are quite different from each other. As aconsequence, form II can be readily distinguished from formI for analytical purposes by determining and comparingtheir respective optical properties. The principal refractiveindices determined for form I are consistent with thosepreviously reported by Jordan.14 The results of a completeoptical characterization for both form I and form II will bepublished elsewhere.The most striking optical difference between the two

polymorphs is seen when they are rotated between crossedpolarizers. Crystals of form II show complete extinctionevery 90° during rotation of the microscope stage. How-ever, when form I is orientated so that it is viewed alongthe b-axis, it shows incomplete and dispersed extinctionwith a distinctive change in interference color from yellowto blue as it passes through the extinction position.As a rapid screening tool, polarized light microscopy was

used to confirm that form II had been isolated in preferenceto form I during the crystallization experiments. Thedispersed extinction displayed by form I was of practicaluse during the experiments to prepare form II fromsolution. This microscopical screening method is verysensitive and it was possible to identify trace amounts ofform I in samples that were analyzed as “polymorphicallypure” form II by powder X-ray diffraction.When gently crushed between a microscope slide and

coverglass, forms I and II fracture in distinctly differentways. Form I usually fractured into irregular fragmentstogether with some polygonal cleavage plates that wereparallel to (010). Form II, on the other hand, gave manythin, diamond-shaped plates that were parallel to (001).

Figure 4sExperimental powder X-ray diffraction pattern for orthorhombicparacetamol (form II).

Table 3sExperimental Powder X-ray Diffraction Data for ParacetamolForms I and II

form I form II

hkl d (Å) °2Θ I/Imax (%) hkl d (Å) °2Θ I/Imax (%)

011 7.301 12.112 26 200 8.562 10.323 4101h 6.398 13.830 18 210 6.935 12.755 5002 5.809 15.240 3 020 5.907 14.986 14101 5.709 15.508 72 211 5.061 17.508 26110 5.640 15.700 4 220 4.862 18.232 22111h 5.294 16.732 11 021 4.618 19.203 49111 4.877 18.175 68 121 4.463 19.878 3020 4.690 18.905 13 400 4.284 20.716 9021 4.355 20.376 39 221 4.066 21.844 22112h 4.276 20.756 7 002 3.699 24.038 100112 3.849 23.087 9 102 3.616 24.597 14121h 3.785 23.483 62 230 3.579 24.855 9022 3.650 24.367 100 420/112 3.464 25.700 13103h 3.596 24.741 5 131/202 3.401 26.177 5122h 3.355 26.545 62 231 3.222 27.663 12211h 3.280 27.165 11 022 3.139 28.407 3202h 3.201 27.847 4 122 3.084 28.931 21122 3.140 28.403 2 312 3.006 29.700 5211 3.075 29.012 7 222 2.945 30.325 27113 3.048 29.273 6 600 2.857 31.280 5023 2.987 29.887 3 240 2.792 32.026 7123h 2.858 31.273 5 322 2.747 32.572 6221h 2.805 31.877 3 141/431 2.706 33.081 4032/131 2.747 32.575 17 132 2.662 33.639 2132h 2.622 34.173 2 611 2.600 34.462 5132 2.513 35.706 3 232 2.573 34.840 7114 2.480 36.193 10 422 2.530 35.454 3033 2.434 36.904 18 512 2.461 36.486 3204h/223h 2.398 37.466 7 440/621 2.431 36.955 17213 2.372 37.896 4 630 2.314 38.891 7

Figure 5sDSC thermogram for monoclinic paracetamol (form I).

Figure 6sDSC thermogram for orthorhombic paracetamol (form II).

688 / Journal of Pharmaceutical SciencesVol. 87, No. 6, June 1998

Crystal MorphologysFigures 7 and 8 show the ob-served morphologies compared with the predicted BFDHand attachment energy models for forms I and II, respec-tively. The drawings of observed morphology were con-structed and indexed by modifying the predicted modelsto match the habits of the crystals that were grownexperimentally.Form IsForm I crystallizes from IMS with a prismatic

to platy habit that is elongated in the direction of the c-axis,but parallel with the {101h} faces. Mature crystals (i.e. thelarger ones that had been growing for the longest time)show the development of the pinacoids {101h} and {101}and the prisms {011} and {110} as the dominant forms.Some immature crystals also show pinacoids {001} andprisms {111}, which are the fast growing faces. Themorphology of form I, as observed by scanning electronmicroscopy, is shown in Figure 9.There is reasonably good agreement between the ob-

served morphology of form I and its predicted attachmentenergy model. However, there is less similarity betweenthe observed and predicted BFDH morphologies. The

BFDH model tends to over emphasize the {110} faces andunder emphasizes the {111h} faces.Form IIsForm II crystallizes from IMS as prisms that

are elongated along the c-axis. The dominant forms inmature crystals are prisms {210} and bipyramids {211}.Immature crystals sometimes show the pinacoids {100},{010}, and {001} which are the fast growing faces. Themorphology of form II, as observed by scanning electronmicroscopy, is shown in Figure 10.In contrast to form I, the observed morphology of form

II shows poor agreement with both the BFDH and attach-ment energy models. The attachment energy model pre-dicts an equant, squat prismatic habit, while the BFDHmodel predicts a habit that is slightly elongated along thec-axis direction. For both of the predicted models, the{111}, {200}, and {020} faces are dominant, whereas theobserved morphology is dominated by the {210} and {211}faces.The dissimilarity between the observed morphology and

the predicted morphologies for form II is in contrast to thesimilarity between the observed and predicted morpholo-

Figure 7sComparison of the observed morphology (grown from supersaturated IMS solution) and the predicted morphologies (BFDH and attachment energymodels) for monoclinic paracetamol (form I).

Figure 8sComparison of the observed morphology (grown from supersaturated IMS solution) and the predicted morphologies (BFDH and attachment energymodels) for orthorhombic paracetamol (form II).

Journal of Pharmaceutical Sciences / 689Vol. 87, No. 6, June 1998

gies for form I. One reason for the dissimilarity shown byform II could be the effect of solvent, which is known tohave a profound influence on the shapes of paracetamolcrystals.21 However, the most likely reason for the differ-ence between observed and predicted morphologies is dueto the rate of crystal growth (probably due to the degree ofsupersaturation) during the preparation of the two poly-morphs. Differences in the morphology of form I, due tothe effect of supersaturation during crystallization, havealso been observed by other workers.22,23 Form I was grownslowly over a period of several hours, so it had time toachieve a dynamic equilibrium with its solution, whereasform II was precipitated rapidly and, as a result, thecrystals did not achieve a dynamic equilibrium. Thissuggests that the observed morphology of form II is aconsequence of its rapid growth rate. Form II crystalsgrown under less stressful conditions may adopt a mor-phology that is more like the equant, squat prismatic habitthat is predicted by the attachment energy model. How-ever, as this present investigation has shown, the isolationof form II from a supersaturated solution relies upon rapidgrowth and so, fully developed crystals such as thosepredicted may rarely be observed using the crystallizationconditions described.

Compaction StudiessThe results of the compactionexperiments are given in Table 4. The yield pressurevalues for form I are in reasonable agreement with a rangeof literature values as given from 79 to 127 MPa.24 Whencompared with form I, the data for form II suggests that ithas a low yield pressure, which is indicative of plasticdeformation. Also, a comparison of the strain rate sensi-tivity values for forms I and II (which describe the per-centage increase in yield pressure over the two punchvelocities24) indicates that form II has the greater increasein yield pressure. This increase is consistent with plasticdeformation, so of the two polymorphs, form II should besuitable for direct compression. After compaction, thepellets of form I disintegrated, while those made from formII remained intact and have not crumbled, even afterstorage for several months.As mentioned in the Introduction, a significant com-

mercial advantage may be gained by using orthorhombicparacetamol in preference to the monoclinic form. This isbecause it undergoes plastic deformation upon compactionand could be directly compressed into tablets without theneed for binding agents. Form I, the commercial form, isunsuitable for direct compression because it does notdeform plastically but breaks by brittle fracture and,therefore, tablets require binding agents.Figure 11 is a photomicrograph showing some crystals

of form II (that are immersed in silicone oil between amicroscope slide and coverglass) that have been fracturedby applying pressure to the top of the coverglass with asteel needle. Note that one crystal shows multiple cleav-ages perpendicular to its length (i.e. parallel with {001}).The observation of these cleavages confirms the existenceof a well-developed slip system which is attributed to theexistence of two-dimensional molecular sheets that lie inthe plane containing the a- and b-axes, as identified by

Figure 9sScanning electronmicrograph showing the crystal habit of parac-etamol (form I) grown from supersaturated IMS solution. Scale bar ) 100µm.

Figure 10sScanning electronmicrograph showing the crystal habit ofparacetamol (form II) grown from supersaturated IMS solution. Scale bar )10 µm.

Table 4sResults from the Compaction Simulator ExperimentsConducted on Paracetamol Forms I and II with Punch Velocities of 2and 220 mm s-1

sample1st run(MPa)

2nd run(MPa)

3rd run(MPa)

mean(MPa)

standarddeviation

strain ratesensitivity (%)

2 mm s-1

form I 70.06 72.30 72.92 71.76 1.23form II 48.42 43.67 42.52 44.87 2.55

220 mm s-1

form I 76.59 79.28 79.88 78.58 1.43 9.5form II 53.87 56.53 56.87 55.76 1.34 24.3

Figure 11sPhotomicrograph showing pressure-induced multiple cleavagesalong slip planes in a prismatic crystal (to the right of center) of orthorhombicparacetamol (form II). Scale bar ) 50 µm.

690 / Journal of Pharmaceutical SciencesVol. 87, No. 6, June 1998

Haisa et al.1 Figure 12 is a molecular diagram (constructedusing data from this investigation) showing these sheetsin the direction of the c-axis and they are held together byhydrogen bonds. Adjacent sheets are weakly bonded (thereare no intersheet H-bonds) and it is because of this thatform II has slip planes and can undergo plastic deforma-tion.

DiscussionThis paper has reported that the growth of the ortho-

rhombic polymorph of paracetamol from solution is arelatively simple procedure. Indeed, 3-5 g of polymorphi-cally pure form II (as determined by PXRD) have beengrown in less than 1 h. So, why has the crystallization ofform II from solution been so elusive since 1974 whenHaisa et al.1 first grew it and determined its structure? Asthis investigation has demonstrated, it is a matter ofharvesting the crystals from the mother liquor soon afterthe crystals have begun to grow. Indeed, if the crystals ofform II are left in contact with the mother liquor for toolong, they will convert to the more stable polymorph, formI.Throughout this investigation, observations were made

about the thermodynamic stability of form II relative toform I. These observations are reported below.Solution-Phase Conversion of Form II to Form

IsThe first indication that form II can be grown fromsolution was the microscopical observation of a solutionphase polymorphic conversion of form II to form I. Duringthe early stages of this investigation, light microscopy wasused to study and gain an understanding of the crystal-lization process of paracetamol at room temperature.Single drops of a saturated solution of paracetamol in

benzyl alcohol were examined after they had been seededwith form II powder. It was noted that the first crystalsto grow were needles of form II. However, within 15 minat room temperature, prismatic and platy crystals of formI began to grow as the needles of form II dissolved. Figure13 shows this solution-phase conversion with two photo-micrographs of the same field of view, taken 30 min apart.Note that the crystals of form I increase in size at theexpense of the form II crystals.Later in the investigation, the same polymorphic trans-

formation process was noted in larger volumes of solutionsof paracetamol in IMS. Depending upon the time allowedfor the growth of crystals once the precipitation hasoccurred, either pure form II, pure form I, or a mixture of

the two will result. These observations indicated that thecrystallization of paracetamol from solution had undergonea solvent-mediated transformation in accordance withOstwald’s law of stages.25A nonseeded, supersaturated solution in IMS held at-75

°C in a stoppered flask took about 7 days to begin tocrystallize. After 21 days, the crystals were confirmed asbeing form II using polarized light microscopy. However,after 43 days at -75 °C, the form II crystals had convertedto form I.An experiment was performed to monitor the conversion

of form II to form I in an IMS solution as a function oftime at 0 °C. Form II was prepared in a flask using theseeding method described previously and polarized lightmicroscopy was used to examine the crystals at specifictime intervals. The solution was gently agitated intermit-tently during this experiment. The results are summarizedin Table 5.Therefore, the most likely reason that form II crystallized

from solution has been elusive to so many other researchersis that the precipitates formed have not been harvestedquickly enough. The conversion of form II to form I seemsto be quite rapid and this is probably due to the fact thatthe solubility of form II is similar to that of form I.9 If thecrystals of form II are collected soon after they have formed,

Figure 12sProjection of the crystal structure of form II (down the a-axis)showing the hydrogen-bonded molecular sheets that lie in the plane of the a-and b-axes (H-bonds are shown as dashed lines).

Figure 13sPhotomicrographs showing the solution phase polymorphicconversion of orthorhombic paracetamol (needles) to monoclinic paracetamol(prisms and plates) at room temperature in saturated benzyl alcohol. Micrographa was taken at t ) 0 and b was taken at t ) 30 min. Scale bars ) 250 µm.

Journal of Pharmaceutical Sciences / 691Vol. 87, No. 6, June 1998

they are unlikely to have had chance to transform to formI. By conducting the crystallization at subambient tem-perature (ideally below 5 °C), the conversion rate of II f Iis retarded and the yield and polymorphic purity of theproduct will be enhanced. This is not a unique phenom-enon and similar saturation-temperature-dependent poly-morph interconversions have been reported for othercompounds.26To develop this laboratory-scale crystallization of form

II into a commercially viable industrial-scale process wouldrequire extensive optimization. This optimization shouldaccount for the rapid conversion of form II to form I, sorapid precipitation and rapid harvesting will be essential.Residual SolventsAs described earlier, it is most

important to fully dry the form II crystals because tracesof residual mother liquor can induce a solvent-mediatedpolymorphic transformation during storage. It may tran-spire that bulk production of form II, free from form I, maybe exceedingly difficult if all of the solvent cannot beremoved rapidly by filtration before it begins to convert toform I. This observation emphasizes the importance ofcomplete drying of form II prior to storage to ensurepolymorphic purity.Prolonged Storage at RoomTemperaturesInspection

of the phase diagram for the paracetamol system showsthat at temperatures above about -30 °C, form I is thethermodynamically stable form.4 This implies that formII is metastable at room temperature and, as a conse-quence, could undergo a conversion to form I duringprocessing or storage. However, samples of form II thathave been fully dried and stored in stoppered vials at room-temperature had not converted to form I after 6 months.Effect of CompressionsApproximately 100 mg of form

II was compressed at 5 tons for 1 min in a 13 mm diameterdie-set using a laboratory infrared press. The pellet waspulverized with a spatula and gently ground to a finepowder using an agate pestle and mortar. Analysis bypowder X-ray diffraction showed that it had not convertedto form I.Effect of GrindingsApproximately 100 mg of form II

was vigorously ground to a fine powder with an agate pestleand mortar for 1 min. When compared with the startingmaterial using PXRD, it was found that the grinding hadnot converted it form I.

ConclusionsThis investigation has shown that the elusive ortho-

rhombic polymorph of paracetamol (form II) can be crystal-lized from a supersaturated solution of IMS by nucleationwith seeds of form II. Form II undergoes a solution-phase

conversion to form I which has been monitored using lightmicroscopy. It has been demonstrated that in order toensure that form II is recovered in preference to form I, itis important to conduct the crystallization at a low tem-perature (0 °C) and to harvest the crystals within 1 h afterthe onset of nucleation. In addition, this investigation hashighlighted the value of light microscopy in the early stagesof a crystallization experiment to observe the microscalecrystallization process so that a successful laboratorymethod can be developed.The crystal structures of forms II and I have been deter-

mined at both room temperature and low temperature byX-ray crystallography.27 These analyses confirm the struc-tures that were published in 19741 and 1976,3 respectively,and in addition, the low-temperature experiments improvethe overall accuracy of the earlier determinations. Differ-ences in the crystal structures between forms I and II arealso reflected in their optical characteristics. Therefore,forms I and II can be differentiated by X-ray diffractionand optical crystallography.The thermal behavior of solution grown form II precludes

the use of differential scanning calorimetry as an analyticalmethod to confirm its identity. This is because it slowlyconverts to form I in the solid state and melts as form I.The compaction properties of form II have been deter-

mined, and when compared with those of form I, form IIhas been shown to undergo plastic deformation. Thisconfirms that form II could be directly compressed intotablets. Observations on the relative stability of form IIsuggest that it does not readily convert to form I uponstorage for several months or if it is mechanically stressed.Although form II has been crystallized in small (several

grams) quantities at the laboratory-scale, further studieswould be needed to establish the feasibility of scaling-upthe crystallization for commercial purposes.

AcknowledgmentsThe authors would like to acknowledge the following Pfizer

personnel for their invaluable contributions during this inves-tigation: Chris Dallman and Dave Smith (for the powder X-raydiffraction and differential scanning calorimetry analyses), AndyMills (for the compaction data), and Paul Fagan (for constructivecomments and advice).

References and Notes1. Haisa, M.; Kashino, S.; Maeda, H. The Orthorhombic form

of p-Hydroxyacetanilide. Acta Crystallogr. 1974, B30, 2510-2512.

2. Nichols, G.; Frampton, C. The Isolation and Characterisationof the Orthorhombic Polymorph of Paracetamol. Posterpresented at the April 1996 meeting of the TripartiteResearch Programme for Pharmaceutical Processing at theUniversity of Strathclyde, Glasgow, Scotland.

3. Haisa, M.; Kashino, S.; Kawai, R.; Maeda, H. The Monoclinicform of p- Hydroxyacetanilide. Acta Crystallogr. 1976, B32,1283-1285.

4. Burger, A. Acta Pharm. Technol. 1982, 28 (1), 1-20.5. Di Martino, P.; Conflant, P.; Drache, M.; Huvenne, J.-P.;

Guyot-Hermann, A.-M. Preparation and Physical Charac-terization of forms II and III of Paracetamol. J. Therm. Anal.1997, 48, 447-458.

6. Fachaux, J.-M.; Guyot-Hermann, A.-M.; Guyot, J.-C.; Con-flant, P.; Drache, M.; Veesler, S.; Boistelle, R. Pure Parac-etamol for direct compression Part I. Development of sintered-like crystals of Paracetamol. Powder Technology 1995, 82,123-128.

7. Fachaux, J.-M.; Guyot-Hermann, A.-M.; Guyot, J.-C.; Con-flant, P.; Drache, M.; Veesler, S.; Boistelle, R. Pure Parac-etamol for direct compression Part II. Study of the physico-chemical and mechanical properties of sintered-like crystalsof Paracetamol. Powder Technol. 1995, 82, 129-133.

8. Garekani, H. A.; Ford, J. L.; Rubinstein, M. H.; Rajabi-Siahboomi, A. R. Manipulation of Acetaminophen Crystals

Table 5sSummary of the Microscopical Observations during theSolution-Phase Polymorphic Conversion of Paracetamol Form II toForm I in a Seeded, Supersaturated IMS Solution at 0 °C over aPeriod of 6 h

time fromseeding observations

10 min well-formed prismatic crystals of form II;form I not found

1 h well-formed prismatic crystals of form II,but showing signs of dissolution

2 h form II crystals undergoing extensive dissolution;a few, small, well-formed form I crystals observed

3 h form II crystals are extensively eroded;form I crystals growing larger

4 h well-formed platy form I crystals are dominant;a few ragged form II crystals remain

6 h all crystals are form I

692 / Journal of Pharmaceutical SciencesVol. 87, No. 6, June 1998

for Direct Compression. Pharm. Res. 1996, 13, 9 (Supple-ment), Abstract PT6211, S-209.

9. Di Martino, P.; Guyot-Hermann, A.-M.; Conflant, P.; Drache,M.; Guyot, J.-C. A new pure paracetamol for direct compres-sion: the orthorhombic form. Int. J. Pharm. 1996, 128, 1-8.

10. Sohn, Y. T. Study on the Polymorphism of Acetaminophen.J. Kor. Pharm. Sci. 1990, 20 (2), 97-104.

11. Nath, B. S.; Khalil, S. S. Studies on Paracetamol CrystalsProduced by Solvent Change Method of Crystallization.Indian J. Pharm. Sci. 1984, 46 (3), 106-110.

12. McCrone, W. C. Fusion Methods in Chemical Microscopy;Interscience Publishers: New York, 1957; p 101.

13. Cosier, J.; Glazer, A. M. A nitrogen-gas-stream cryostat forgeneral X-ray diffraction studies. J. Appl. Crystallogr. 1986,19 (2), 105-107.

14. Jordan, D. D. Optical Crystallographic Characteristics ofSome USP Drugs. J. Pharm. Sci. 1993, 82, 1269-1271.

15. Hartshorne, N. H.; Stuart, A. Crystals and the PolarisingMicroscope, 4th ed.; Edward Arnold: London, 1970.

16. Docherty, R. The Application of Computational Chemistryto the Study of Molecular Materials. Conference Proceedings:Crystal Growth of Organic Materials; Myerson, A. S., Green,D. A., Meenan, P., Eds.; American Chemical Society: Wash-ington, DC, 1996; pp 2-14.

17. Heckel, R. W. Density-Pressure Relationships in PowderCompaction. Trans. Metall. Soc. A. I. M. E. 1961a, 221, 671-675.

18. Heckel, R. W. An Analysis of Powder Compaction Phenom-ena. Trans. Metall. Soc. A. I. M. E. 1961b, 221, 1001-1008.

19. Welton, J. M.; McCarthy, G. J. X-ray Powder Data forAcetaminophen. Powder Diffraction. 1988, 3 (2), 102-103.

20. Wilson, C. C.; Shankland, N.; Florence, A. J.; Frampton, C.S. Single-Crystal Neutron Diffraction of Bio-Active SmallMolecules. Physica B (Amsterdam) 1997, 234-236, 84-86.

21. Green, D. A.; Meenan, P. Acetaminophen Crystal Habit:Solvent Effects. Conference Proceedings: Crystal Growth ofOrganic Materials; Myerson, A. S., Green, D. A., Meenan,P., Eds.; American Chemical Society: Washington, DC 1996;pp 78-84.

22. Grant, D. J. W.; Chow, H.-L. Crystal modifications in acet-aminophen by growth from aqueous solutions containingp-acetoxyacetanilide, a synthetic impurity. AlChE Symp. Ser.1991, 87 (284, Part. Des. Cryst.), 33-37.

23. Finnie, S.; Ristic, R. I.; Sherwood, J. N.; Zikic, A. M. Char-acterization of growth behaviour of small paracetamol crys-tals grown from pure solution. Trans IChemE 1996, 74(A),835-838.

24. Rowe, R. C.; Roberts, R. J. Mechanical Properties. In Phar-maceutical Powder Compaction Technology; Alderborn, G.,Nystrom, C., Eds.; Marcel Dekker: New York, 1995; pp 283-322.

25. Cardew, P. T.; Davey, R. J. The kinetics of solvent-mediatedphase transformations. Proc. R. Soc. London 1985, A398,415-428.

26. Khoshkhoo, S.; Anwar, J. Crystallization of polymorphs; theeffect of solvent. J. Phys. D: Appl. Phys. 1993, 26, B90-B93.

27. Details of the crystal structure determinations at 123 K forboth form I and form II can be obtained from the Director,Cambridge Crystallographic Data Centre, 12 Union Road,Cambridge, Cambridgeshire CB2 1EZ, England. Any requestfor this supplementary data should include the full literaturecitation.

JS970483D

Journal of Pharmaceutical Sciences / 693Vol. 87, No. 6, June 1998